计算机应用 ›› 2014, Vol. 34 ›› Issue (12): 3433-3437.

• 人工智能 • 上一篇    下一篇

基于有向双关系图和多核融合的蛋白质功能预测

孟军,刁印   

  1. 大连理工大学 计算机科学与技术学院,辽宁 大连 116024
  • 收稿日期:2014-06-23 修回日期:2014-08-11 出版日期:2014-12-01 发布日期:2014-12-31
  • 通讯作者: 孟军
  • 作者简介:孟军(1964-),女,辽宁大连人,副教授,博士,CCF会员,主要研究方向:机器学习、数据挖掘;刁印(1984-),男,陕西咸阳人,硕士研究生,主要研究方向:机器学习、数据挖掘。
  • 基金资助:

    辽宁省自然科学基金资助项目

Protein function prediction based on directed bi-relational graph and multi-kernel fusion

MENG Jun,DIAO Yin   

  1. School of Computer Science and Technology, Dalian University of Technology, Dalian Liaoning 116024, China
  • Received:2014-06-23 Revised:2014-08-11 Online:2014-12-01 Published:2014-12-31
  • Contact: MENG Jun
  • Supported by:

    Natural Science Foundation of Liaoning Province of China

摘要:

针对多源异构蛋白质相互作用网络信息量大、数据冗余导致预测结果不能充分反映数据分布信息的问题,将功能类别网络和蛋白质相互作用网络相结合,提出基于有向双关系图和多核融合的多标记学习算法。首先,构建基于含有损失函数的目标方程和最大期望算法的自适应模型;然后,利用图优化策略融合功能类别和蛋白质相互作用网络构成的多个关联矩阵;最后,将融合后的关联矩阵代入模型中预测蛋白质功能。在Yeast和Mouse的蛋白质多源异构数据上的实验结果表明,提出的方法具有预测准确率高、标签损失率低等优势。

Abstract:

In view of the problem that protein interaction network of multiple kernels from heterogeneous data sources contains huge amount of information. Due to data redundancy, the predicted results could not fully reflect the distribution of data. The functional categories network and protein interaction networks were combined, a multi-label learning algorithm was proposed based on the directed bi-relational graph theory and multi-kernel fusion. First, an adaptive learning model was built by the loss function of equation and expectation maximization algorithm. Then, multiple associative matrices were obtained by using the graph optimization strategy to fuse the functional categories and protein interaction networks. Finally, the prediction model was built by the associative matrices and adaptive learning model. The experimental results using multiple heterogeneous protein data sources of Yeast and Mouse show that the proposed method has higher prediction accuracy and lower loss rate of label.

中图分类号: